Incident update as an avatar video: text first, then a 30s clip

Do not make video your incident channel: publish the text update first, and add an avatar clip afterwards for customers who prefer to watch. A 30-second Sume avatar video costs about $5.52 at Standard, $7.35 at Plus (the default) and $16.50 at Max with no product image, and it renders as an asynchronous job, so it fits the follow-up, not the first minute.

What a clip adds to a status update

During an outage people read short factual lines. A clip helps in two cases: a long incident where the update changes slowly and you want a calmer, consistent voice, and a post-incident explainer that walks through impact, cause, and what changes. For a live, evolving situation, text is faster to edit and does not go stale.

A script skeleton

Keep it near 28-32 seconds, inside the 4-60 second window Sume accepts, with one fact per sentence. A useful order is: what happened, who is affected, what you have done, what to do now, and when the next update comes. Use plain words and avoid blame or guesses about cause until it is confirmed.

Use multi-scene video_inputs if you want a silent beat between the facts and the action; the scenes share one background and one avatar. Add a first-frame preview so a second person reads the first frame and the script before you pay for the render. Turn on inline captions: many viewers watch without sound, and a failed caption stage soft-fails to the clean video, so check captions.status before posting. See Video captions for styles.

{
  "avatar_handle": "status_host",
  "aspect_ratio": "16:9",
  "quality": "standard",
  "captions": { "enabled": true, "style": "slam", "language": "en" },
  "script": "Since 14:05 UTC, some customers cannot sign in. We have found the cause and a fix is rolling out. You do not need to change anything. Our next update will be at 15:00 UTC on the status page."
}

Operations notes

• Pre-create the avatar once ($0.95) and keep a template script, so the job is a fill-in-the-blanks edit during the incident.
• Render at Standard for a time-sensitive clip. Standard is described in the docs as the fastest Sume execution path, and Max is slower.
• Use aspect_ratio: "16:9" for a status page embed, and 9:16 (the default) for social.
• Keep the text on the page as the source of truth; the clip is a convenience.

Limits and honesty

A render has no guaranteed completion time in the docs, so never promise a video by a specific minute. A script is estimated into a duration, and one over 60 seconds is rejected, which is a good limit for incidents anyway. Label the presenter as an AI avatar, and never put a real person's name or face on a statement they did not make. If the facts change after the render, re-render; an edited script needs a new preview.